Milvus Vector Database

Collect metrics information related to Milvus vector database

Configuration¶

Prerequisites¶

• DataKit is installed
• Install Milvus

Enable Milvus Monitoring¶

Enter the Milvus /home/$USER/milvus/conf directory, modify the server_config.yaml file, and enable Milvus monitoring

metric:
  enable: true
  address: 127.0.0.1 
  port: 9091

Note: If Milvus is deployed in a Docker container, the above address should be changed to the host machine IP.

• After modification, restart Milvus and check the monitoring data via http://localhost:9091/metrics

Configure DataKit¶

• Enter the conf.d/prom directory under the DataKit installation directory, copy prom.conf.sample and name it milvus.conf

cp prom.conf.sample milvus.conf

• Adjust the content of milvus.conf as follows:

## Exporter URLs.
  urls = ["http://127.0.0.1:9091/metrics"]

  ## Stream Size. 
  ## The source stream segmentation size, (defaults to 1).
  ## 0 source stream undivided. 
  stream_size = 1

  ## Unix Domain Socket URL. Using socket to request data when not empty.
  uds_path = ""

  ## Ignore URL request errors.
  ignore_req_err = false

  ## Collector alias.
  source = "milvus"

  ## Measurement prefix.
  ## Add prefix to measurement set name.
  measurement_prefix = ""
  measurement_name = "milvus"

  ## (Optional) Collect interval: (defaults to "30s").
  interval = "30s"

  # [inputs.prom.tags]
    # some_tag = "some_value"
    # more_tag = "some_other_value"

Other configurations can be adjusted as needed, parameter adjustment description:

• urls: Metrics addresses, fill in the corresponding component exposed metrics url here
• source: Collector alias, recommended for differentiation
• keep_exist_metric_name: Keep metric names
• interval: Collection interval
• inputs.prom.tags: Add extra tags

Restart DataKit by executing the following command

datakit service -R

Metrics¶

Milvus Metric Sets¶

Milvus metrics are located under the Milvus metric sets, here we introduce the relevant descriptions of Milvus metrics